Custom DOE Returning Rsquared of 1

GSheehan1

Hello All,

 

I am running a custom design DOE with 3 factors at 2 levels each but limited to only 10 runs. One of my factors is a hard to change variable. I have 4 responses in my DOE, 1 rated with importance of 1, 2 with importance of 2 and the remaining item with importance of 3. As you will see from the attached file, when the model script is run I get an RSquare value of 1 for "Delta in ROn Mean from t0->t1" and "Delta in VtOn Mean from t0->t1" responses which I can't make sense of.

 

I am quite new to DOE and this is my first time using a custom design and hard to change variable. Is an Rsquare value of 1 expected due to the fixed effect tests? Should I be focusing more on the t ratio for each variable to determine which are the most important factors rather than the RSquare and Prob>t? Thank you.

JMP Custom Design DOE File_New3.jmp

statman

I did not have a chance to look at your dataset, but thanks for posting it.  My first guess is an RSquare of 1 simply means your model is saturated.  That is, every available degree of freedom is assigned a model effect.  As you remove terms from the model, you will get an RSquare and an RSquare adjusted as well as p-values etc.  There are a number of methods used to determine significance of unreplicated designs: Normal plots, Pareto plots, Bayes plots, etc.